摘要
在采用机器视觉测量零件尺寸的过程中,提出了一种基于局部区域灰度矩图像边缘定位方法。该方法先利用中值滤波对采集到的图像进行处理,然后通过Prewitt算子对边缘进行初步定位,在此基础上,对图像边缘附近的9个像素的灰度值进行一维灰度矩定位,最后采用误差分析和处理的方法,获得更精确的边缘位置。实验证明,该方法能有效提高边缘检测精度,在检测分辨率为1000 DPI情况下,尺寸检测误差小于2μm。
An edge positioning method of local area gray moment is presented in research of using machine vision to measure the size of parts. Firstly, median filtering is used to process the image. Secondly, image edge of part by Prewitt operator is located approximatively. Then, close to the edge, nine-pixel gray is used for positioning with one-dimensional gray moment. Finally, error analysis and treatment method are used for a higher precision location on the edge. The experiment results show that in the circumstances of 1 000 DPI detection resolution, the accuracy for measuring diameter could reach to 2um.
出处
《光学与光电技术》
2009年第3期60-63,共4页
Optics & Optoelectronic Technology
基金
广东省高等学校自然科学重点研究(05Z010)资助项目
关键词
视觉检测
尺寸检测
边缘检测
灰度矩
亚像素
visual inspection
size detection
edge detection
gray moments
sub-pixel
